1962

179

IRE TRANSACTIONS ON MILITARY ELECTRONICS

An RF Multiple Beam-Forming Technique*

WILLIAM P. DELANEYt, MEMBER, IRE
Summary-An RF beam-forming matrix is described which forms
"tn" simultaneous beams from an "n" element array in a passive and
theoretically lossless manner. The principle of operation is explained
using some simple matrix configuration. A general expression for
the far-field pattern of any beam is derived and then used to study
the positions of beam peaks, the position of beam nulls, the crossover
level between beams and the frequency sensitivity of beam positions.
This matrix provides a uniform illumination of the array aperture;
however, simple beam combining techniques will yield tapered il
luminations. An experimental 16-element beam-forming matrix
which operates at 900 Mcps is described, and results of RF and
antenna measurements on the matrix are presented.

I. INTRODUCTION

flFPHE PHASED ARRAY radar is often proposed as a

solution to many present and future radar problems.
All phased array receivers combine the outputs of
the discrete antenna elements to form one or more antenna
beams. This combining process can take place either in
the IF portion or the RF portion of the receiver. One major

and successfully tested at Lincoln Laboratory. The results

presented below are applicable to any size beam-forming
matrix. The experimental data presented is a condensa.

tion of extensive test results on a 16-element matrix.

II. PRINCIPLE OF OPERATION
The basic components of the beam-forming matrix are
3-db directional couplers or hybrid rings and fixed phase
. .
.
shifters. Before explaining the operation of the matrix, it
is necessary to adopt some conventions concerning the
phase shift through hybrids and directional couplers. The
conventions used are shown in Fig. 1. When the input
voltages have the amplitudes and relative phase angles
shown in Fig. 1, all the input signal power will appear at
the indicated terminal.

-9-O ,v v,v' IvL/-90- Iv/z

disadvantage of beam forming in the lower-frequency por-

tions of the receiver is the need for amplitude and phase

stable mixing and amplifying circuits between the antenna
and the LF beam-former. This paper describes a theoreti-

* Received by the PGMIL, November 30, 1961. The work reported

was performed at Lincoln Laboratory, Lexington, Mass., with support
from the U. S. Air Force.
t M.I.T. Lincoln Laboratory, Lexington, Mass.
1

J. Blass, "Multidirectional antenna-a new approach to stacked

beams, 1960 IRE INTERNATIONAL CONV1ENTION RECORD, Pt. 1, PP.. 48m," 1

2J.

Butler and R. Lowe, "Beam forming matrix simplifies design of

electronically scanned antennas," Electronic Design, vol. 9, pp. 170-

173; April 12, 1961.

3 J. P. Shelton and K. S. Kelleher, "Multiple beams from linear

arrays," IRE TRANS. ON ANTENNAS AND PROPAGATION, vol. AP-9, pp.

154-161; March, 1961.
4J. Butler, "Multiple Beam Antenna," Sanders Associates, Inc.,
Nashua, N. H., Internal Memorandum RF-3849; January 8, 1960.

5J. L. Allen, et al., "Phased Array Radar Studies, 1 July 1959 to

1 July 1960," M.I.T. Lincoln Lab., Lexington, Mass., Tech. Rept. No7

228;rAugu,Mst,[Link] Do. No7. 297,HyeLiry,M.

Iv!

L|
Ov 1/iv/-90

/2v/-90 Ov

cally lossless multiple beam-forming technique which

operates directly at the RF carrier frequency and thus can

be located directly behind the antenna elements. All the
beams formed by this technique have the full gain of the
array aperture. Since the beam-forming technique uses a
matrix of passive microwave devices, it can be made very
rugged, reliable, and is not susceptible to the phase and
amplitude instabilities common in active RF devices.
Simultaneous RF multiple beam-forming techniques are
relatively new. Blass' has described a series-fed multiple
beam antenna which is presently used in the AHSR-1 airport radar system. Butler and Lowe2 and Shelton and
Kelleher' have described the parallel-fed technique discussed in this paper. Experimental results on a 4-element
matrix4 at 3 Gc have been reported by Butler. Shortly
afterwards, an 8-element matrix' at 900 Mcps was built

Iv/0 Iv/-180-

/ivZi

s/2n:

Ov Ov

HYBRID

Fig. 1-Phase-shift conventions for directional

couplers and hybrid rings.

.--- -INCIDENT WAVEFRONTS

-9O

0.

-l
-90

0.

7ANTENNAS

B
BEAM

LEFT

BEAM

RIGHT

Fig. 2-Simplest beam-forming matrix.

Now it is possible to form the simplest multibeam array
by using two antenna elements and one hybrid ring or one
3-db directional coupler. Fig. 2 shows a 2-beam, 2-element

array using a 3-db directional coupler. A certain incident

wavefront will excite antenna element currents that are 900
out of phase, and therefore all the received signal energy
will come out one terminal on the directional coupler. Thus,
a "beam right" and a "beam left" are formed. If a hybrid

ring had been used, the 2-element array would have a

broadside beam and an endflre beam (assuming an antenna
lmn saigcoe ooehl wvlnt)
eeetsalgcoet n afwvlnt)
A 4-beamn matrix can be built by interlacing two 2-beam
matrices arid then providing a second level of directional
couplers or hybrid rings to combine the outputs into beams.
It is necessary to insert fixed phase shifters between the

IRE TRANSACTIONS ON MILITARY ELECTRONICS

180

upper and lower levels of couplers to form the output beam.

Fig. 3(a) shows a 4-element, 4-beam array using directional couplers. The amplitudes and phases of an incident
"beam 1 Left" signal are shown at various points in the
matrix. Fig. 3(b) shows the amplitudes and phases of an
incident "beam 2 Left" signal. The "shaded" directional
couplers are the ones used to form the particular beam.
Thus, it can be seen that the beam-forming matrix behaves
like a multiple corporate feed structure which routes a
signal originating at a particular point in space to a particular output port of the matrix. Since the matrix is symmetrical about a vertical line through the center, the 1
Right and 2 Right beams can be found by using the approach of Fig. 3.
A 16-element matrix can be thought of as four interlaced 4-element matrices. Two extra levels of phase shifters
and combining elements (directional couplers or hybrid
rings) are required to form the beams. Fig. 4 shows the
diagram of the 16-element matrix used to obtain the experimental measurements presented in this paper.
This beam-forming technique can be used in two-dimensional (planar) arrays by first combining the columns of
antenna elements in matrices and then combining the outputs of the column matrices in a group of row matrices.
At this point, it is worthwhile to list and briefly discuss
some of the fundamental characteristics of the RF beamforming matrix. Some of these topics are discussed at
greater length later.
Number of Beams: The number of beams formed is
equal to the number of antenna elements used. The

DIRI;CTIONAL

COUPLER

FIXED -4-

April

_Nc,DENTWAvEFRoNr
__

.vl

ov

____

///

SHIFTER

0. l

2V/-I8O,

0'

BEAMI LEF

(a)

OEH .,AvEF_2L \~

IVL@

____

2,,

Iv/-05

Iv/-270
2

AA

2v/-135

oV

BEAM RIGHT

Ov

BEAM 2 LEFT

BEAM 2 RIGHT

Ov
BEAM LEFT

(b)
Fig. 3-(a) Amplitude and phases of a "beam 1 Left" signal in a 4element matrix. (b) Amplitude and phases of a "beam 2 Left" signal

in a 4-element matrix.

ANTENNAS
2

1-76.75'l

0'

o0'

-11.25'

-3375'

0'@

0
|-O

IL

BR

5L

4R

*3L

6R

7L

10

-56.25'

-5625'

0'

5*

3375*

2R

2L

7R

6L

3R

12

BEAM TERMINALS

Fig. 4-A 16-element beam-forming matrix.

i3

-*15.24L

14

15

0'

0'7-78.5*

5R

8L

16

Delaney: RF Multiple Beam-Forming Technique

1962

181

number of antenna elements in the array must be equal to antenna element number 1 also connects to antenna
to a power of 2, namely 2, 4, 8, 16, 32, etc.
element number 9. Similarly, elements 2 and 10 are fed by
Number of Directional Couplers or Hybrids: The num- a common directional coupler. The first beam from broadber of combining devices required can be shown to equal side is formed when an incident wavefront excites currents
N/2 log2 N where N==number of elements in the array. in elements 1 and 9 (and thus elements 2 and 10, 3 and 11
Number of Fixed Phase Shifters: The number of phase etc.) with a phase difference of 900. Similarly the second
shifters required is N/2 (log2N- 1).
beam from broadside is formed when this phase difference
Operating Frequency: The operating frequency is is 2700, and in general the Kth7 beam is formed when the
limited only by the practicality of building and inter- phase difference between elements 1 and 9 is (2K- l)r/2
connecting fixed phase shifters and directional couplers radians. Therefore, for a 16-element array, the element to
or hybrids. (Hybrid transformers can be used in VLF element phase shift for the kth beam is (2K-1)ir/16
matrices.)
radians.
Bandwidth: The basic components (phase shifters and
Generalizing this result to an "n" element array,
directional couplers) can be built with bandwidths
(2K - 1)ir
greater than 30 per cent; however, there are certain
K==(3)
n
problems involved in operating phased arrays over
bandwidths of this size.
Combining this result with (1) and (2), the normalized farInsertion Loss: The beam-forming technique is field amplitude of the Kth beam from broadside is
theoretically lossless and in practice matrices can be
built with low values of insertion loss.
1 r
-rd
2K
s
Antenna Array Illumination: The matrix provides a
1 ssn
n
n-ns/
uniform illumination of the array elements. However,
EK = - _
(4)
n
sird nsimple beam combining techniques can yield (cosine)n
illuminations.
LX
n 12]
The "antenna" characteristics of this matrix such as where:
beam shape, beam position, beam crossover, etc. are dis|EKI =normalized (peak = 1) magnitude of far-field field
cussed in Section III.
intensity
n=number of elements in the array
III. THEORETICAL PERFORMANCE IN AN
d= element to element spacing
ANTENNA SYSTEM
AS= wavelength
The normalized magnitude of the field intensity in the
K =beam number
far field of a linear array of n isotropic sources is given6 by:
a= angle from the array normal.
If

sin-

(1)

beam-forming matrix

field amplitude of the

sin-

|Ep'2ird

(2)

built using hybrid rings

as

Pth5 beam can be shown to be

.

where

were

combining devices instead of directional couplers, the far-

1
n

sin n

Fird .
L

Pir]
-sisn P1

.frd.
sin -sin

a-

Pir]

Hereafter this paper will consider only matrices using 3-db

d =element spacing
directional
couplers. Similar results can be obtained for
X = wavelength
hybrid ring matrices by using (5).
a=angle from the array normal
Eq. (4) is now used to examine various antenna properties
=-element to element phase [Link]-rmnmtix
Of the beam-forming matrix.
The element to element phase shift, b, can be found by

referring to the schematic diagram of the particular matrix

K is defined as the beam number. Values of K are limited to positive
under consideration. Using a 16-element array as an ex- nonzero integers. When these values of K are used, the equations which
thrfr
[Link]
followthe
apply
onlyforms
to beams
the right
of the symmetrically
array normal. about
However,
ample, anand~~~~~~
matrix
hrfr
eern otemtl ceal fsince
are located
b)eamstowshich
the
Fig. 4, it is seen that the directional coupler which connects array normal, this restriction does not hinder the usefulness of the results.
6 J. D. Kraus, "Antennas," McGraw-Hill Book Co., Inc., New York,
N. Y., p. 78; 1950.

The mathematical simplicity more than compensates for the slight loss
in generality.
8 Positive values of P yield beams to right of broadside. Negative
values of P yield beams to left of broadside. P=O yields the broadside
beam.

182

IRE TRANSACTIONS ON MILITARY ELECTRONICS

Positions of Beam Peaks

The peaks of various beams occur at values of the angle
ca which make the numerator and denominator of (4) equal
zero. Designating these angles as "ap"

sin ap

1I

nd

( nq + K--)
2

(6)

where q = any integer.

A value of q = 0 yields the position of the main beam.

Nonzero values of q give the positions of grating lobes.

Positions of Beam Nulls

Pattern nulls occur at those zeros of the numerator of
(4) which are not accompanied by zeros in the denominator. Designating angles of pattern nulls as ao,
x /

(m+K-Knd\
2/7
where m = 1, 2, 3, etc. but m X qn
where q = any integer.
sinao=

Beam Crossovers
The angular position at which two beams cross over can
be found9 from (4) by equating the amplitudes of the Kth
and the adjacent (K+l)th beams. Defining "a," as the

angular position of

the crossover,

KI,

a, = sin-1

(10)

The beam crossover level E| is found by substituting (10)

into (4):
1
E1

sin

For values of n> 10

E

(7)

Relative Positions of Peaks and Nulls

April

~~~~~~~~~~~~~~~~~~~~2
2
(12)

Thus the crossover level is independent of beam position,

element spacing and wavelength and for larger arrays
(n> 10) it is effectively constant at 2/r. Since the peak of
any beamn has an amplitude of unity due to the normalizaof (4), this crossover level is approximately -4 db
(-3.92 db).

P
RElative
oftPeak
Eqs. (6) andns
(7) point
outandtNulstingcharacteristicstion
two interesting characteristics
7)point

of beams formed by this technique:

1) All the nulls of all the beams occur at certain common

angles.
2) The peak of any beam occurs at one of these same
common angles.

Thrizes

characteristics of the matrix beams are

in Fig. 5 which shows three typical beams.
BEAM tIBEA9 2

sum-

BEAM 3

g~~~~~~ivenby

I/f

Beam Movement with Changes in Frequency

As the operating frequency is changed the position of the
peak of any matrix beam changes due to the factor of X
in (6). The amount of beam movement is also a function of
beam position as evidenced by the beam number, K, in
(6). This fact will complicate any compensation scheme
designied to allow operation over wide instantaneous bandwidths. The change10 in beam position with frequency is

dap

., |\

= tan

ap

dX

(13)

For beams near endfire where acp approaches

differential equation (13) can be solved to give:
Sector Coverage
The angular sector 6, covered by the multiple beams from
the matrix, can be found from (6). "6" is the angle between
the peaks of the extreme right and extreme left beams
(K=n/2 for these beams).
62

[n
([

)]

(8)

-A 2V2/_X
X
ap

900,

the

(14)

As an example consider the effect of a 10 per cent frequency

change. The near-in beam positions will be shifted by
about 10 per cent; whereas a beam close to endfire will
have its position shifted by 28 per cent.
IV. TAPERED APERTURE [Link]
a uniform illumination of b)eam-*forming
the andtenna array. While this illumination provides
The

For large where

arrays n>~~~1

6~~
2sn1 \)

matrix provides

the narrowest beamwidth and the greatest directivity, it

determinled

Thus for a large array the sector coverage is

only by the element spacing and the wavelength,

9The solution of interest occurs when ,6= -{K+i where ^6 is defined

in (1).
10 Assume the phase shifters and the couplers in the matrix are insensitive to changes in frequency.

Delaney: RF Multiple Beam-Forming Technique

1962

has a relatively high first sidelobe level (-13 db). Lower

sidelobes can be achieved at the expense of beamwidth
and directivity by combining the output beams of the
matrix. Allen" has shown that it is possible to achieve
arbitrary aperture illuminations using weighted addition
of selected beams. The discussion below considers the
simple addition of two adjacent beams.
The addition (after phase correction) 12 of two adjacent
beams will result in a cosine illumination of the aperture.
This fact is readily demonstrated by adding the illuminations of the Kth and the (K-+ 1)th beams. In the interest of
mathematical simplicity the array is considered as a line
source for this derivation.

IK

Ej(2K-1)#

K
IK+1 = Ej(2K+1)

where

IK = illumination of Kth beam

7r

3=-

2 l
I = array length
x = length variable along the array.
'K + 'K+r1 = eJ(IK i)2 E2(IK+i):
= Ei2K#[E-' + E+Hi]
= (2 cos 3)Ei2K#.

183

the uniform illumination beam. The crossover point between the adjacent cosine beams is down 2.1 db from the
peak.
There are some limitations on forming adjacent taperedillumination beams as shown in Fig. 6. Allen'3 has shown
that the beam shape and beam spacing obtainable from a
lossless simultaneous beam-forming matrix are not completely arbitrary. The results of this analysis indicate that
if the beam-forming process is to be lossless the crossover
level of adjacent tapered illumination beams cannot be
higher than the crossover level of the uniform illumination beams. Thus, the cosine illumination beams of Fig. 6
cannnot be realized simultaneously in a lossless manner.
This fact is easily recognized if one tries to build a beam
combiner to simultaneously form the beams of Fig. 6.
Signal energy from the center uniform illumination beam
must be split to form the two cosine beams. Therefore,
when a signal arrives on the peak of the left cosine beam,
some of the signal energy is coupled to the right cosine
beam and thus is lost in the beam combining process.
Cosine illumination beams which are formed simultaneouly and losslessly are located further apart than the beams
of Fig. 6. The crossover levels of such beams are 9.5 db
down from their peaks.
An approach similar to the one used to develop (15) will
show that three uniform illumination beams can be combined to form a cosine-squared illumination of the array.
Also, by a proper choice of how these beams are combined
it is possible to achieve cosine-squared-on-a-pedestal
illuminations.
Section V presents experimental data taken on a multiple
beam-forming system. Antenna patterns for uniform,
cosine and cosine-squared illuminations are included.

(15)
Thus the amplitude of the new illumination is a cosine
function, and the phase distribution of the new illumination points a beam halfway between the K and the K+1
component beams. Fig. 6 shows the addition of uniform
illumination beams to form cosine illumination beams.
A 1/V/2 factor is introduced in the addition process to keep
V. EXPERIMENTAL RESULTS ON A 16-ELEMENT
BEAM-FORMING MATRIX
the power in a cosine illumination beam the same as the
power in a uniform illumination beam so that the gain of
matrix
in
g
the beams can be compared. Fig. 6 shows that the cosine the shematic diagramofthis
beam has 0.92 db less gain than the uniform illumination No attempt was made to achieve an especially compact
beam and a null-to-null beamwidth 1.5 times greater than
package in this experimental model. The use of strip
transmission line permitted the fabrication of several di-----IR -R -3R
1.0---;_rectional couplers and phase shifters on the same board.
The boards were then interconnected using coaxial fittings.
2/7,
/ / \ /\
- -

1+2Re
1/,/-2- 0.707

The directional couplers14 and phase shifters'5 used in the

t . - /matrix are of the coupled strip type. Both these com-ponents have bandwidths greater than 30 per cent. These
wideband components were used to more fully explore the
SR_ __possibilities of this beam-forming technique. The measurements reported here were made at 900 Mcps.

"7
",
\.

<

Fig. 6-Formation of cosine illumination beams.

13 J. L. Allen, "A theoretical limitation on the formation of lossless

multiple beams," IRE TRANS. ON ANTENNAS AND PROPAGATION, vol.

"Allen, et al., op. cit., pp. 221-223.

12 Fixed phase shifters must be inserted in the beams to guarantee inphase addition of the signal energy. For an example of a phase correction
calculation see W. P. Delaney, "An RF Multiple Beam Forming
Technique," Lincoln Lab., Lexington, Mass., Group Rept. No. 41G-0012;
Sec. 4.1; 1961. (ASTIA No. 262 017).

AP-9, pp. 350-353; July, 1961.

14 J. K. Shimizu, "A Strip Line 3 db Directional Coupler," Stanford
Rtes. Inst., Stanford, Calif., SRI Project 1592, Sci. Rept. No. 1 (AF19
604-1571); June, 1957.
11 B. M. Shiffman, "A new class of broad-band microwave 90-degree
phase shifters," IRE ON MICROWAVE THEORY AND TECHNIQUES, vol.
MTT-6, pp. 232-238; April, 1958.

IRE TRANSACTIONS ON MILITARY ELECTRONICS

184

April

Antenna Measurements

'~The beam-forming matrix

was tested in a 16-element

linear array of parallel dipoles spaced 0.58X apart. Fig. 8
shows the major lobes of all sixteen beams. The angular
, sector covered by all the beams is 1070 which agrees closely
' with the value of 1080 predicted by (9). The beam peaks
occur very close to the positions predicted by (6). The
E worst case is a 1.20 error in beam 7 Right. The envelope of
the beam peaks follows the antenna element pattern very
closely as would be expected from the principle of pattern

multiplication.
Figs. 9-13 show typical antenna patterns of individual
matrix beams. The sidelobe performance of the near-in
beams (1 Right, 2 Left, 3 Right) is quite good with first
sidelobes of - 13 db. The antenna element spacing of
0.58X used in this test allows grating lobes to form for

beams further than 450 off broadside. The start of this

grating lobe structure is evident in beam 7 Right (Fig. 13).
It was shown earlier that a cosine illumination of the
aperture could be achieved by adding two adjacent beams.
RF Measurements at 900 Mcps
Fig. 14 shows a cosine illumination beam obtained by addThis beam-forming matrix is a matched device at all in- ing beams 4 Right and 5 Right in a hybrid ring. The first
put and output ports. The maximum VSWR at any port sidelobe in this beam is about 22 db down which is close to
the theoretical value of -23.2 db for a cosine illumination
is 1.27.
Any beam terminal is isolated from any other beam on a 16-element array. The hybrid ring also provides the
terminal due to the inherent isolation of the directional difference of the two adjacent beams (corresponding to a
couplers. The average isolation at 900 Mcps is 28 db with a sine illumination). The difference pattern, Fig. 15, has a
lowest value of 15 db. The antenna terminals of the matrix sharp null at the same position as the peak of the sum beam
are isolated from each other in the same manner. The aver- and thus, it may be useful for amplitude comparison monopulse applications.
age isolation is 30 db with a lowest value of 20 db.
Fig. 16 shows the antenna pattern of a cosine-squaredThe insertion loss of the matrix is 0.74 db. Practically
all of this insertion loss is due to the strip transmission on-a-pedestal illumination. Weighted addition of beams 1
line which has a loss of 0.18 db per foot at 900 Mcps. The Left, 2 Left and 3 Left yielded a 24 per cent pedestal
use of new strip-line materials and more compact packag- (theoretical highest sidelobe= -27 db). The first sidelobes
ing should allow a significant reduction in this insertion measure -27 db but amplitude and phase errors in the
component beams and in the beam combiner cause a few
loss.
The matrix provides a uniform amplitude illumination other sidelobes to exceed this level. The peak of the beam
of the array. However, slight variations in the coupling occurs at the peak of the 2 Left beam as expected.
VI. CONCLUSION
ratios of the directional couplers cause amplitude errors
in the illumination. The worst rms amplitude error is 0.41
In conclusion, it is worthwhile to reiterate the addb and the average rms error for all beams is 0.30 db.
RMS phase errors in the array illumination average 4.80 vantages of this beam-forming technique and also point out
p
for all beams with a worst case of 12.90.
Fig. 7--The 16-element matrix.

~~~~~~~~~~~~00 4
Fig. 8-Major lobes of at sixteen beams (vertial scale=40 db,

horizontal scale=20

per_smallest division).

horizontal
scall.,lile =2' per smaillillles

ivision).

Fig. I9-Beam
Right...erticats ...........[Link]

scale=20 per smallest..ivision).

cal=

pr sallst(liisin

Delaney:

1962
Jill ltill

11 11
11 11

II IIIII 111111
111111 111111
ill IIIII

H II
Jill
Ill

II III IIII III 11 III

IIIIIII III 11 IIIIIII 11 11
Ill II 111111I 11 Ill
H

III

III H II IIIIIII

IIIIIIIIIIIIII
H-H.

H.,

Vil 11

11.

I11- I It

Jill

ill 11

IIII II

IIIIIIII
lifilill

Ill

.
t-

dlill 111
Ill
tillilill

IH IIIIII

IIIII Illill
1111 lfilli

11

IIIIII

itill
IIIII

lllu 1

11

11

Ad

uuiit
.19 1111

fit

Ill

JIM

11

10-Beam 2 Left.

H 11

1 I
IIIII

H ill
IL
H ill
IIH Ill

Ill
Ill

ill,

IN
n iiiiiiiiii

IIII
fill

IllL

H ill

Jill

111

1111

Ill

1111111111

Jill.
11 11

Jill

H li
II II
1111 Jill
IIII Ji ll

H II
111111
ll

IIII

Jill

Ill]
Jill
Jill
ilH
IlH

IIIIIIIIIIIIIII
II IIII IIIIII II
11111
ill
''I

11

fit

IIIII
Jill

1111 It _LLLL
Ill It lilt
Ill ILM

-1411

H IM

IIIIII I

III II

11

_tLL
Iii
IIll

Ill

...

.......
.... 11111

itill Ill
Ill

LLI
LL

11111111

IIIIIII

_1 IIIIIII

II

1111 11

HH

IIIIII

Ill

H IM
III
H IS 'I Jil
1111
liizi
lie
111

114 11
lih

IIII

11111 1111111

111 H IH II

11
ill H ill
IIIII

11-Beam 3

III

111111

IIIIIIIIII

11

1111

11 II If

IIII

IIIII 11

Ill

IIIII H

Right.
ll
f
if
IIIIIIIIIIIIIII 111111111111111 Ill 11111 1111
IIH 11 11111111111 11 Ill 11111 11
ll II
11
H
Lilli
11111
ill
11
II

111111

111

11111

JIM

H
H ill
[I ll

It

ill

Iiiiii

IIIIII
111

11

1111

LA.

Ill
11

IIIIIIII

IIIIIIIII.-

II

Ill
Ill
Ill

Ill

Ill

liLli

Ill.

flill

AL 11- 1

Iffl ill

11

lfill

111111

11

H ill
111111

II II
H ill
H ill

Ill

11 ill 11 11
11 H it
11 1111

IIII IIIII

1111

fill 11

(Ji ll

lfilill
ifilill
11 1111

4 11

Jill

H it 11
ll 1111 11
lilt till

ill

Jill

11A

-fi 11111 Ill

LH-LL

44

II. IIIIIII II
111111L
II II III

...

IIII1-i

111111

Ill]

-I

111-T

11

11 H it Ill

1- 1

ifillillilill
illillillil

II

ll

H li
IIIIII
IIIIIII '1111
1111111 11111i
II IIIIIIIIIIIII

1111 IIIIIII

Ill

[Ill IIIH III

...I
I.,
Ill
Ill
filill

lillillit

A-M

lf

IIIII

litill

1_11
.11111
41LL

11111

r-

1111

11

IIII Hi ll
Jill lm il
H lig fit
1111 ll
ill

111111111
III 111
H it IIII
ill
1111 11

ill IIII

_H L

_LHM
IIH IM
IA LLI-LL
LLLLL

Fig.

111111

IIIIIIII

IIA III

Ill

IIIIIII

IIIII

11111
TM
11111

II IIIIII IIIIII
IIIIIIIIIIII

Iiii

II11

_LL

itil

Ilitill

-L-

1111111
iiLlit

1111L

lfili

III

IIIIIIIII
11 IIIIIIIII
11 IIII II III

IIII

[N il

11
IIIIIIIII 11

11
11 lfilil

11 L il

Il

11I I11

til V L_

=L

Fig.

til A A IL

itilill

Ill

IL U
ill 11 iiiiiin
ll 11 IIIIIIH

IIIII

ll

IIIIIH
1111111

Ill

Ill

1 11111

11

illi

Ill

H ILL

11

H IM

1 IIII
H ill

IIILIEIII
ill
ll

till

Ill

IIIII
H ill
H ill

Ill

11111

II

I..

123
N

IIIII

Hill

HUM

Jill

IIH IIII

IIIIIIIIII

Ill

Iiiii

V111111111 LILL

IIIIIIIIII
IL -L

ilillillillitil

fill
Illi
Jill
Jill

t5i

......

IIII H IIII

LLUT II 9

..

H ill. H ill

fill IIIIII

111111 11-

Ill 1111H -M
IllIHIM 1111
[Link] lititL litil
II w

.... 111.

LM
Ill

IIIIIIIIIII

Ill]]

IIIU IIIIII

IHIM

ri III

II

1111

11 1-1

IIIIII H

L Ill
r

111111

Jill
fi llilli t il if 11111 111
11- ifill Ill
I11
11 II
Ill
H ill Ill 11 I'll

1, 1111111 11 11 11 II IIIIII

- IT 111111111
111111111

IllH

H[Hill
11

185

Multiple Beam-Forming Technique

RF

liki Ill

11 H ill

.1. 1.

1,

1. 11

11 kil

Ill-

''I

11

11

ill
I.

H il ill 11 11

11

lff

111 1111

Ul

-EaLra llituttillit niti IIIII

niiiii 11 IIIIIII
Ill il

E11H.1 = 11
L AM

Fig.
IIIIIIIIIII
1111111
IIIIIII
1111111

11

HIM
litill
111111
ill
H IM

1111.1111 11 11 111

Jill iiin iiiii

Jill
H li
H li
1111

LLH-tll t
111111
JIM
11111
IIIII
IIIII

if illill illi
IIIII
litill JIM

-1 IIII
J IlH
I Jill

lfilli 111.11

fu ll Ill

fill

IIIII Ill
ill Ill

ill H ill ill

Ill 11111

ill H IIH
Ill Ill

II

Jill Ill H ill Ji ll ill IIII

Jill Ill Jill Jill
IIII IIII .1111
H IIIIIIII II

till

ill lill
IIIIII III If

IIIIIIIII

IH IIIIII 11 II 111111111 1.

111

III II III 11

If

ifilifill 11

1111 _111 I

.111filiniii.

ILLI EL

litililill

lfml

Fig.

Illillill
Iiii u

Ill LM

IIIIIII II

IIIII II II

H ill11

4ill if
IU A

III

IIL-LL

ill IIIII Ill

itil IIIII 11
Ill] IIIII
ill

it _H

H ill Jill
111111 Ill H lH H li
111111 Ill IIII
IIIII

H li

lifil
11111
IIIII

IIIIII
IIIIII

[Ill

H li
IIII
IIII

11

11111
_LLLW

Jill

__111

ill

IIIII IIIIIIIIII

IIIIIIIII

111111111
fillit
lifill
[11111
Lti ti

LLittl

H IM
H IM
IIIIII

Ill IIIII
Ill IIIII
ill tlill Ill Ill H ill
Ill IIIII Ill IIIII H ill
IIII H ill
11111111

111111

H IM

iim

IIIII

11

IIII
111
11 11

II

H it
ill

ifill
11111
JIM

]ILL

11
11
ill
Ill
Il

i till
I
AllI
I

11
11

II
]

11

11
IIIII
ill
ill

Jill
Jill
111 11
H ll

JIM

11

L ___LL

LU

H IM IIII

H
11

1111 1114, 1

11111

ill
ill

IIII IIIII.
IlH IIIII
Jill Ill

1111 ill
ill

IIIIIH
H IM

ill

ill
II.

Ill

ill II

11

II. 11

11

iiiin iiii II
llilLilill 11

I
111M

111H

ill

Ill
111111

IIIIII
IIIIII

ii iiiti iiH
illi lilill

litLi
L H

IlH

11
H ill

JIM

IIIIII

IIIIIIIII

+1
H ill,!
H 111

i lilL

IIIIH ff

ill IlH

11

11
H

II Mt
Ali
liff

illiLl
iiiinI

1111,111

Ill

Ill

IIIIIIIII
litilli

W ill

ill
il

IR

Ill

liiii iiiii n ii

H ill

H IU

Jill
ill

III III III Ji ll Jill 111111

IIIIIIIIII IIII Jill IIIIII
1,111111111 H IM IIIIII

11111

'fill H ill Ill

Ill IIIII Ill

lifill

LLLLH

HHH

Jill
Jill

H IM
I

TT[TT=

IIRIII

1111u.111 Jill

Right.

13-Beam 7

111111 11 fl ill II II II II II Jill III III III II fil

H li III IIIIII H
if ill II III III

IIIIII

lilt
Jill fill
II IIII Jill

III II IIII Jill

IIIIIIIII

Fig.
II IIIII Ji ll
III IIIIII JIM
IIIIIIIII Ili]

ivii

Hl

12-Beam 5 Left.

ill 1111111 11 ill 11 II IIIIII II IIII IIIIIIIII IIII

IIIIIIIII Ill 11 IIIIII ill H IH H IM ill H ill
ill
111111111 111, 11 lfilli Ill
II. 1111 Ill HIIIII
Li A
IIIIIIIII Ill
IIIIII Ill
N Ill Ill
111111111 Ill
111111 Ill
lfil ill Ill
H IM Ill
.11
filitilli Ill

.111 11111 ill

III

1ULMILL

tiLl.
E.1 1liffTflT

-ILL H

ill

11 iA lti
tillillill

H ill
IIIII

LLLLL

IL

= T7T
mi t

iTm

14-Cosine illumination beam (beam 4

Right plus beam 5 Right).

IiIIII I11111 111

tilli lli
IIIIIIII

11

Fig. 15-Sine illumination beam (beam 4 Right minus beam 5 Right).

III III IIIIII III lf illill ililli

1111 H
111,1111111111 H IM
41
11111
H 111 111111
H ill
Ill Ill
H II
III Ill
11
rli
H
II
11 111
IIIII II
il
H
11
IIIIIIIIIIIIII
IH IIIII
[it
H
-LE11.1 11
11.1

]Zllll[LLI
I

11 H ll
11 IIII
11
11 II
II

Jil IIII
Ill 11 11
il 11
11
ill
il
II

11 (IIIIII IL

H il fill
IIII H li
IIII 1111
1111 IlH

H ill
11 11
11 11
11 If

II II IIII IIIIII
II IIIIIIIIIII

IIIII
1111
IlH
Jill

I! IIIIIIIII11IIIIIIIII

II..

IIIIIIIIII

iiiiiiiaiiii,

411111111111'

186

IRE TRANSACTIONS ON MILITARY ELECTRONICS

Considered from an antenna standpoint, the major advantage of this technique is the realization of simultaneous
multiple beams, all of which have the full gain of the aperture. Rather than having to steer a single beam using, for
example, RF phase shifters, one has only to observe the
outputs of the matrix in a selective or simultaneous manner. In view of the continuing advances in microwave
switching diodes, it seems reasonable to reduce the number
of receivers required by time sharing a particular receiver
to several matrix beams using diode routing switches.
Since the matrix is theoretically lossless (and in practice has a low insertion loss) it can, in many applications, be
located directly behind the antenna elements with a resultant saving in phase and gain stable HF circuits. The
microwave elements used in the matrix are passive and
nonvariable; thus the matrix can be made rugged and
reliable using strip transmission line techniques.
The matrix itself can be made broad-band, but the
frequency limitations of phased arrays (as opposed to
time-delayed arrays) are still present.
A major drawback of this beam-forming technique, when
large arrays are considered, is the complexity of the matrix
arrangement. For example, a 64-element matrix requires
192 directional couplers and 160 fixed phase shifters. Thus
some ingenuous fabrication and packaging techniques are
required if low matrix insertion loss is to be maintained.
A possible drawback of this technique from an antenna
standpoint is that the simultaneous beams are fixed in
space. An angle tracking technique such as null tracking

April

monopulse would have to be replaced by amplitude com-

parison monopulse.

The matrix is limited in power-handling capability

(from a transmitting standpoint) by the high power characteristics of the transmission line used. For large matrices
it is discouraging to think of using coaxial line or waveguide
and thus the power limit is set by the strip transmission
line. Also, since the matrix behaves like a power divider,
the bottom directional couplers and phase shifters will
have to handle much higher powers than the upper ones.
For lossless simultaneous beam forming the aperture
illuminations are limited to a uniform illumination which
has relatively high first sidelobes or tapered illuminations
which have far-field patterns with undesirably low crossover levels.
Several of these problem areas such as matrix fabrication and packaging, angle tracking with fixed beams and
high-power operation of strip transmission line are presently under study.

The author wishes to acknowledge the contributions of

Dr. Judd Blass, J. Butler and P. Shelton to multiple beamforming techniques. The 16-element matrix described in
this paper was built by Sanders Associates.
The author is especially indebted to J. Allen, J. Resnick
and J. Sklar of Lincoln Laboratory for their helpful discussions on beam forming.